Martin (1992: 114-5):
The final set of options to be developed is also associated with relevance phoricity — but of a comparative rather than superlative kind. Comparative reference differences [sic] from superset reference in that it combines freely with both presenting and presuming groups: a/the larger frog, a/the more attractive option. Presenting groups on the other hand do not presume supersets in English. In apparent counterexamples to this principle such as [3:37] the superlative adjective functions as a Classifier, not an Epithet, and does not presume a group of prizes of which the participant in question is the second (similarly A First Book in German, i.e. 'a primer'):
[3:37] What's this one then?— It's just a second prize I picked up at the Easter Show.
 To be clear, Martin's notion of 'presenting and presuming groups' arises from mistaking nominal group deixis for cohesive reference. In Martin's examples, the 'presenting groups' are those with non-specific deixis (a) whereas the 'presuming groups' are those with specific deixis (the). In SFL theory, reference is made by reference items, not nominal groups, and whereas the specific determiner functions as a (demonstrative) reference item, the non-specific determiner, by definition, does not. Further, it is the items larger and more attractive that function as instances of comparative reference.
 Given that Martin distinguishes presenting from presuming, on his own terms, it should be no surprise that presenting groups do not presume.
 To be clear, second is not a 'superlative adjective', but an ordinal numeral functioning as Classifier.
 This again confuses construing experience (as prizes, one of which for second place) with textual reference. Moreover, even in its own terms, it is the direct opposite of what is true, since identifying a prize for second place does indeed presume a set of prizes that includes at least one other member: a prize for first place.